AI avatars

The source material your AI avatar is built from, filmed with precision

An avatar model learns exactly what it is given: every flicker in the light, every shift in the framing, every echo in the voice. We film the material it learns from to the standard the platforms ask for, so what you get back is as close to you as the tool can make it.

A presenter filmed for an AI avatar on a studio set at Estudio Helios, Medellín

Precision is the whole job

01

Light that never moves

Flicker-free studio light, set once and locked. The model sees the same skin tone, the same shadows and the same exposure from the first second to the last.

02

A camera that never drifts

Sony cinema cameras on locked-off supports, focus and white balance fixed by hand. Nothing auto-adjusts mid-take for the model to learn by mistake.

03

Sound with nothing in it but you

A Shure SM7B in a treated room, recorded separately from the camera. No echo, no hum and no room tone for a voice clone to copy.

04

Direction for the machine

Scripts written to cover the sounds and expressions the models need, and a filmmaker who directs pace, eyeline, hands and stillness for the model rather than for an audience.

Choose by what you want the AI to do

An AI Set example: a presenter at a desk on a dark set with a lamp, greenery and a branded screen, at Estudio Helios, Medellín

AI Set

USD 490includes 90 minutes in the studio, or +USD 200 on a VIP pass

You, on one of our sets, ready to be replicated.

A regular studio session on the set you choose, filmed to a script designed for avatar training. Your avatar platform learns you in that set, so every video it makes afterwards appears filmed in the same place, in the same light, in the same outfit.

Best for

Talking-head avatar platforms such as HeyGen and Synthesia: course updates, explainers, translations and short videos without a new shoot.

What you get

  • One look on the set of your choice, filmed from our avatar script
  • Horizontal and vertical masters, plus a platform-ready copy
  • A clean voice file for voice cloning
  • Your consent clip, recorded on the day
  • Reference stills
  • Extra looks (a change of outfit or set) at USD 150 each
Book an AI Set
A presenter filmed against a plain studio backdrop for an AI clone at Estudio Helios, Medellín

AI Clone

USD 980includes four hours in the studio, or +USD 400 on a VIP pass

A complete record of how you look, move and sound.

A full capture session, built for tools that recreate you anywhere rather than in one place: video on green screen, and a systematic photo set on a plain neutral backdrop, every side of you in even, controlled light. The tool learns your face and your proportions, not a room.

Best for

Identity and character tools such as Higgsfield that place a lifelike version of you in any scene, and professional voice cloning. It also works on talking-head platforms, though for a presenter in one place the AI Set looks more natural.

What you get

  • Green screen video from the front, both three-quarter sides and full length, lit for a clean key
  • A front-on speaking master from our clone script, plus listening and idle takes, turns and a walk as movement reference
  • A systematic photo set on a neutral backdrop: front, three-quarter and profile, close, medium and full length, neutral and expressive
  • A curated training selection sized to your tool (Higgsfield takes 20 to 80), plus every frame from the shoot
  • 30 minutes of clean read speech, the minimum for professional voice cloning
  • Your consent clip, recorded on the day
Book an AI Clone
Side by sideAI SetAI Clone
Filmed onA studio set of your choiceGreen screen for video, a neutral backdrop for photos
Built forTalking-head avatars (HeyGen, Synthesia)Lifelike clones in any scene (Higgsfield) and professional voice cloning
VideoOne locked-off angle, horizontal and verticalFront, both three-quarters and full length
PhotographyReference stillsA full photo set from every side on a neutral backdrop, with a curated training selection
VoiceA clean voice file (enough for instant voice cloning)30 minutes for professional voice cloning
Studio time90 minutes includedFour hours included; the session runs three to four
DeliveryWithin three working daysWithin five working days
PriceUSD 490USD 980

An avatar is a tool, not a stand-in for you

These tools are improving quickly, and their results vary: between platforms, between versions of the same platform, and from one script to the next. Some outputs pass at a glance; others drift in the eyes, the hands or the voice. We control the quality of what goes in. No studio can control what a platform makes of it, so nobody can honestly promise you a result.

No avatar replicates real human presence. People decide whether to trust you from small things: timing, breath, eye contact, the way you react. A clone gets close to how you look and sound. It does not carry what you mean.

Used carelessly, an avatar works against the brand it was made for. An uncanny video under your name, an avatar where a client expected you, words in your voice you would never say: credibility takes years to build and one clip to dent.

Where an avatar earns its place

  • Translations into languages you do not speak
  • Updating a course or onboarding video when one detail changes
  • FAQs, product walkthroughs and internal training
  • Testing hooks and scripts before you film them for real

Where to show up yourself

  • Your story, your opinions and anything said in the first person
  • Apologies, difficult news and anything sensitive
  • Sales conversations and first impressions with a new client
  • Anything your audience is paying to hear from you

Tell your audience when a video is AI-generated wherever it matters to them, and keep your consent clip and source files private. The best results we see come from people who use their avatar for the routine and save themselves for the moments that count.

FAQs

The AI Set if you want a talking-head avatar that presents from one recognisable place, the way it would in a course or an explainer. The AI Clone if you want the tool to place you anywhere, generate photos and video of you in new scenes, or clone your voice to a professional standard. If all you need is a presenter in HeyGen or Synthesia, choose the AI Set: a real set looks more natural than a replaced background, especially around hair.

The AI Set is filmed to the common requirements of talking-head platforms such as HeyGen and Synthesia. The AI Clone covers those too, and adds what identity and character tools such as Higgsfield train on: a varied, well-lit photo set with full-length frames, and multi-angle video. The voice recordings suit voice cloning tools such as ElevenLabs. If your platform publishes specific guidelines, send them before the session and we follow them.

So the tool learns you and nothing else. A set teaches the model a room as well as a person, which is what you want for the AI Set and the opposite of what a clone needs. The green is lit separately, and you stand far enough from it that none spills onto skin or hair, so the key is clean and your colour is true. The photos are taken on a neutral backdrop instead: training sets work best with clean, consistent light and no colour cast on the skin.

It is the minimum ElevenLabs accepts for a professional voice clone, and it gives a convincing result. ElevenLabs accepts up to three hours, and more clean audio gives a closer match, especially for long-form narration. Extra voice time can be added to the session as studio time.

No, and you should be wary of anyone who does. The result depends on the platform, its current version and how it is used. What we guarantee is the material: filmed and recorded to the platforms' published requirements, consistent from start to finish, and delivered on time.

No. We film, photograph, record and deliver the source material. The avatar or clone is created on the platform of your choice, on your own account, by you or your team. That keeps your likeness in your hands.

For the AI Set, what you would wear in the videos the avatar will make: a plain mid-tone top, no fine stripes or small checks, no reflective jewellery. For the AI Clone, nothing green and nothing shiny, fitted clothes that show your build, and your hair as you normally wear it. For both: no hats, sunglasses or scarves, and glasses only if you always wear them. Keep the same look all session so the model learns one consistent you.

No. We provide the avatar script for the AI Set and the clone script for the AI Clone, both written to cover the sounds, expressions and movements the models learn from. You read from the XL teleprompter. If your platform supplies its own script, bring that instead.

You do. Everything is delivered to your Google Drive. The studio never uploads your footage, photos or voice to any AI platform, and never uses them to train anything.

The AI Set is USD 490 with 90 minutes in the studio; on a VIP pass the studio time comes from your hours and the package is USD 200. The AI Clone is USD 980 with four hours in the studio, or USD 400 on a VIP pass. Choose AI Set (it appears as AI Avatar Base Footage) or AI Clone as the format when you book. Colour grading and audio optimisation are included in both. Extra time for other content in the same session is billed at the usual hourly rate.

Start with the material, and the tools can only get better

Platforms change every few months. Footage filmed properly today is still good source material for the tools of next year.